Scott R. Moisik

A study of laryngeal gestures in Mandarin citation tones using simultaneous laryngoscopy and laryngeal ultrasound (SLLUS)

In this work, Mandarin tone production is examined using simultaneous laryngoscopy and laryngeal ultrasound (SLLUS). Laryngoscopy is used to obtain information about laryngeal state, and laryngeal ultrasound is used to quantify changes in larynx height. With this methodology, several observations are made concerning the production of Mandarin tone in citation form. Two production strategies are attested for low tone production: (i) larynx lowering and (ii) larynx raising with laryngeal constriction. Another finding is that the larynx rises continually during level tone production, which is interpreted as a means to compensate for declining subglottal pressure. In general, we argue that larynx height plays a supportive role in facilitating f0 change under circumstances where intrinsic mechanisms for f0 control are insufficient to reach tonal targets due to vocal fold inertia. Activation of the laryngeal constrictor can be used to achieve low tone targets through mechanical adjustment to vocal fold dynamics. We conclude that extra-glottal laryngeal mechanisms play important roles in facilitating the production of tone targets and should be integrated into the contemporary articulatory model of tone production.

A high-speed laryngoscopic investigation of aryepiglottic trilling

Six aryepiglottic trills with varied laryngeal parameters were recorded using high-speed laryngoscopy to investigate the nature of the oscillatory behavior of the upper margin of the epilaryngeal tube. Image analysis techniques were applied to extract data about the patterns of aryepiglottic fold oscillation, with a focus on the oscillatory frequencies of the folds. The acoustic impact of aryepiglottic trilling is also considered, along with possible interactions between the aryepiglottic vibration and vocal fold vibration during the voiced trill. Overall, aryepiglottic trilling is deemed to be correctly labeled as a trill in phonetic terms, while also acting as a means to alter the quality of voicing to be auditorily harsh. In terms of its characterization, aryepiglottic vibration is considerably irregular, but it shows indications of contributing quasi-harmonic excitation of the vocal tract, particularly noticeable under conditions of glottal voicelessness. Aryepiglottic vibrations appear to be largely independent of glottal vibration in terms of oscillatory frequency but can be increased in frequency by increasing overall laryngeal constriction. There is evidence that aryepiglottic vibration induces an alternating vocal fold vibration pattern. It is concluded that aryepiglottic trilling, like ventricular phonation, should be regarded as a complex, if highly irregular, sound source.

Modeling the Biomechanical Influence of Epilaryngeal Stricture on the Vocal Folds: A Low-Dimensional Model of Vocal-Ventricular Fold Coupling.

PURPOSE Physiological and phonetic studies suggest that, at moderate levels of epilaryngeal stricture, the ventricular folds impinge upon the vocal folds and influence their dynamical behavior, which is thought to be responsible for constricted laryngeal sounds. In this work, the authors examine this hypothesis through biomechanical modeling. METHOD The dynamical response of a low-dimensional, lumped-element model of the vocal folds under the influence of vocal-ventricular fold coupling was evaluated. The model was assessed for F0 and cover-mass phase difference. Case studies of simulations of different constricted phonation types and of glottal stop illustrate various additional aspects of model performance. RESULTS Simulated vocal-ventricular fold coupling lowers F0 and perturbs the mucosal wave. It also appears to reinforce irregular patterns of oscillation, and it can enhance laryngeal closure in glottal stop production. CONCLUSION The effects of simulated vocal-ventricular fold coupling are consistent with sounds, such as creaky voice, harsh voice, and glottal stop, that have been observed to involve epilaryngeal stricture and apparent contact between the vocal folds and ventricular folds. This supports the view that vocal-ventricular fold coupling is important in the vibratory dynamics of such sounds and, furthermore, suggests that these sounds may intrinsically require epilaryngeal stricture.

Harsh Voice Quality and Its Association with Blackness in Popular American Media

Performers use various laryngeal settings to create voices for characters and personas they portray. Although some research demonstrates the sociophonetic associations of laryngeal voice quality, few studies have documented or examined the role of harsh voice quality, particularly with vibration of the epilaryngeal structures (growling). This article qualitatively examines phonetic properties of vocal performances in a corpus of popular American media and evaluates the association of voice qualities in these performances with representations of social identity and stereotype. In several cases, contrasting laryngeal states create sociophonetic contrast, and harsh voice quality is paired with the portrayal of racial stereotypes of black people. These cases indicate exaggerated emotional states and are associated with yelling/shouting modes of expression. Overall, however, the functioning of harsh voice quality as it occurs in the data is broader and may involve aggressive posturing, comedic inversion of aggressiveness, vocal pathology, and vocal homage.

Quantal biomechanics in an embodied phonetics

Quantal regions were described by Stevens [e.g., 1989, J. Phon. 17, 3–45] to identify nonlinear stabilities in the relationship between articulation and acoustics. Classic cases of quantal effects show how tongue posture may vary within one region of the vocal tract with little acoustic change, while in other regions very small movements can have large effects on acoustic output. Such effects can be thought of as attractors to speech behavior in those regions of the phonetic space that allow greater noise. Quantal-like stabilities have been suggested to operate not just in articulatory-acoustic space, but in biomechanical-articulatory space as well [e.g., Schwartz et al., 1997, J. Phon. 25, 255–286]. It is argued here that such quantal-like stabilities are a hallmark of speech modules [Gick & Stavness, 2013, Front. Psych. 4, 977], providing the basis for robust, feed-forward control. Computer simulations in the ArtiSynth platform (www.artisynth.org) are used to demonstrate quantal effects in speech biomec...

Agent model reveals the influence of vocal tract anatomy on speech during ontogeny and glossogeny

This study introduces a new model for the investigation of the complex manner in which vocal tract anatomy affects human speech production and may influence language change and evolution. The anatomy of the human vocal tract has long been recognized to play a crucial role in speech production and patterning (Fant, 1971; Ohala, 1983). It imposes discrete relations between articulatory parameters and acoustics (Stevens & Keyser, 2010), with highly nonlinear mappings between them (Stevens, 1968, 1989), and it has been recently suggested that inter-individual and patterned inter-population variation in the anatomy of the vocal tract might play a role in explaining patterns of linguistic diversity (Dediu, Janssen, & Moisik, 2017). We investigate these complex relationships by instructing a computersimulated agent to learn to reproduce, as well as possible, target speech sounds by controlling the articulators of a detailed 3D geometric model of the human vocal tract based on the VocalTractLab 2.1 (Birkholz, Jackèl, & Kroger, 2006), modified to allow changes in larynx height and hard palate shape. More precisely, the agent minimizes the Euclidean distance (in the F1–F5 formant space) between the target and the produced sounds using a genetic algorithm that optimizes the synaptic weights of a neural network that maps formants to articulatory parameter values1. Here, we apply this model to two case studies, both using the five-vowel system [a], [æ], [i], [u], and [@], but investigating the effects of variation in different components of the vocal tract. In the first case study, we revisit the debate concerning the role of larynx height in human speech, which has important implications for the evolution of speech

Modelling human hard palate shape with Bézier curves.

People vary at most levels, from the molecular to the cognitive, and the shape of the hard palate (the bony roof of the mouth) is no exception. The patterns of variation in the hard palate are important for the forensic sciences and (palaeo)anthropology, and might also play a role in speech production, both in pathological cases and normal variation. Here we describe a method based on Bézier curves, whose main aim is to generate possible shapes of the hard palate in humans for use in computer simulations of speech production and language evolution. Moreover, our method can also capture existing patterns of variation using few and easy-to-interpret parameters, and fits actual data obtained from MRI traces very well with as little as two or three free parameters. When compared to the widely-used Principal Component Analysis (PCA), our method fits actual data slightly worse for the same number of degrees of freedom. However, it is much better at generating new shapes without requiring a calibration sample, its parameters have clearer interpretations, and their ranges are grounded in geometrical considerations.

Talking heads: Morphological variation in the human mandible over the last 500 years in the Netherlands

The primary aim of this paper is to assess patterns of morphological variation in the mandible to investigate changes during the last 500 years in the Netherlands. Three-dimensional geometric morphometrics is used on data collected from adults from three populations living in the Netherlands during three time-periods. Two of these samples come from Dutch archaeological sites (Alkmaar, 1484-1574, n=37; and Middenbeemster, 1829-1866, n=51) and were digitized using a 3D laser scanner. The third is a modern sample obtained from MRI scans of 34 modern Dutch individuals. Differences between mandibles are dominated by size. Significant differences in size are found among samples, with on average, males from Alkmaar having the largest mandibles and females from Middenbeemster having the smallest. The results are possibly linked to a softening of the diet, due to a combination of differences in food types and food processing that occurred between these time-periods. Differences in shape are most noticeable between males from Alkmaar and Middenbeemster. Shape differences between males and females are concentrated in the symphysis and ramus, which is mostly the consequence of sexual dimorphism. The relevance of this research is a better understanding of the anatomical variation of the mandible that can occur over an evolutionarily short time, as well as supporting research that has shown plasticity of the mandibular form related to diet and food processing. This plasticity of form must be taken into account in phylogenetic research and when the mandible is used in sex estimation of skeletons.

FRANK: A Hybrid 3D Biomechanical Model of the Head and Neck

Abstract We describe our approach to construct FRANK: a Functional Reference ANatomical Knowledge (FRANK) template of the head and neck. FRANK consists of a collection of anatomical components made of finite element models (FEM), rigid-bodies, spring-like structures, and various muscle types, all wrapped by an airtight parametrically controlled geometric covering. Its muscles can be activated to mimic complex actions such as swallowing, chewing, and speech, and its modular design allows for components to be replaced and tailored to target-specific applications. The underlying biomechanical modeling toolkit, ArtiSynth, uses a hybrid finite-element and multibody technique that is essential in allowing simulation of such processes within reasonable computation times. We also describe common challenges and various approaches to building hybrid models, such as FRANK, enabling the combination of organs into a unified framework. When combined with inverse modeling techniques, these hybrid models can be applied to a wide variety of applications, including modeling of function. We demonstrate some applications of FRANK, including swallowing, mastication, speech, and patient-specific anatomical modeling.

Glottal squeaks in VC sequences

This paper reports results related to the phenomenon referred to as a “glottal squeak” (coined by [1]). At present, nothing is known about the conditioning and the articulation of this feature of speech. Our qualitative acoustic analyses of the conditioning of squeaks (their frequency of occurrence, duration, and f0) found in Aberystwyth English and Manchester English suggest that squeaking may be a result of intrinsically tense vocal fold state associated with thyroarytenoid (TA) muscle recruitment [2] required for epilaryngeal constriction and vocal-ventricular fold contact (VVFC) needed to produce glottalisation [3]. In this interpretation, we hypothesise that squeaks occasionally occur during constriction disengagement: at the point when VVFC suddenly releases but the TAs have not yet fully relaxed. Extralinguistic conditioning identified in this study corroborates findings reported by [1]: females are more prone to squeaking and the phenomenon is individual-dependent.